MODELING LONGITUDINAL AND DYADIC DATA WITH HLM
<p>Aline Sayer, associate professor of psychology and CRF's Methodology Program Director at UMass Amherst</p>
The hierarchical linear model (HLM) provides a conceptual framework and a flexible set of analytic tools to study a variety of social, political, and developmental processes. One major application focuses on the modeling of longitudinal data where time series data are clustered within persons. A second application concerns the analysis of dyads, where individual responses are clustered within couples, sibships, caregiving dyads or other matched pairs. This workshop will consider the formulation of statistical models for these two applications using the HLM6 software package.
The workshop will provide an introduction to the basic two-level model for polynomial growth functions, splines (piecewise growth models), checking model assumptions, multiparameter hypothesis testing, the incorporation of time-varying covariates, and multivariate models for growth, with consideration of a variety of alternative covariance structures. Learning the computing necessary to fit the models is an integral part of the course.
After a brief review of older methods for analyzing dyads, the workshop will focus specifically on the HLM modeling approach for analyzing dyadic data. These include dyadic consensus and discrepancy, multivariate outcomes, and the actor-partner interdependence model. Special models for exchangeable dyads (such as identical twins or same-sex friendship pairs) and the extension to longitudinal models will also be presented.
The course will meet 8 hours per day, with equal time devoted to lecture-demonstration and a computer lab using the HLM program. Participants should have strong backgrounds in multiple regression analysis.
return to main Methodology Program and Consulting Services